The desire to look attractive is universal. Researches show that 55% of first impressions of others on every one of us depend on our dressing. Thus, dressing plays a very important role in our daily life. However, if feature information (such as clothes categories and clothes colors) of clothes can be learned in advance when the clothes are selected, it will be advantageous for us to select fitted and favorite clothes.
In the prior art, when feature information of clothes is recognized, feature information recognition is performed on the overall pictures containing clothes pictures. In most cases, the overall pictures contain background pictures, and the clothes pictures only account for a part of the overall pictures. Therefore, when feature information recognition is performed on the overall pictures containing the clothes pictures, the accuracy and precision of recognition may be reduced, and even a situation that a recognition result is completely wrong occurs.